Cardiovascular Disease Risk Prediction

A machine learning model and web application to predict cardiovascular disease risk.

---

📌 Table of Contents

• Project Overview
• Key Features
• Installation
• Usage
• Data
• Model Performance
• Deployment
• Contributing
• License

---

🎯 Project Overview

Objective: Predict cardiovascular disease (CVD) risk using clinical, demographic, and lifestyle data.
Target Audience: Clinicians, researchers, and individuals seeking proactive health insights.

Components:

1. Data Analysis: EDA, outlier handling, and feature engineering.
2. Predictive Model: XGBoost classifier (73.67% test accuracy).
3. Web Application: Streamlit app for real-time risk prediction and data exploration.

---

🔑 Key Features

• Engineered Features:
  • BMI categories (Underweight, Normal, Overweight, Obese).
  • lifestyle_score (smoking, alcohol, inactivity).
  • Blood pressure categories (Normal, Elevated, Hypertension Stage 1/2).
• Model: XGBoost with 15 key clinical biomarkers (e.g., age, cholesterol, blood pressure).
• App Functionality:
  • Real-time risk prediction with probability scores.
  • Interactive visualizations (univariate, bivariate, multivariate analysis).

---

⚙️ Installation

1. Clone the repository:

   git clone https://github.com/yourusername/cvd-risk-prediction.git  
   cd cvd-risk-prediction  

Install dependencies:

bash pip install -r requirements.txt
Required Libraries:

Python 3.8+

pandas, NumPy, scikit-learn, XGBoost

Streamlit, Matplotlib, Seaborn

🚀 Usage Run the Streamlit App: bash streamlit run app.py
Input Fields:

Demographics: Age, gender, height, weight.

Health Metrics: Blood pressure, cholesterol, glucose.

Lifestyle: Smoking, alcohol, physical activity.

Output:

Risk Level: "High Risk" or "Low Risk" with a confidence score.

Dashboard: Explore data trends via interactive plots.

📊 Data Dataset: cardio_data_processed.csv (68,205 records, 14+ features). Key Variables:

Target: cardio (1 = CVD present, 0 = absent).

Features: age_years, ap_hi, ap_lo, bmi, cholesterol, lifestyle_score, etc.

Preprocessing:

Outliers in height, weight, and blood pressure were replaced with medians.

Categorical variables encoded for modeling.

📈 Model Performance Best Model: XGBoost

Metric Value Test Accuracy 73.67% F1-Score 0.74 Precision 0.74 Recall 0.74 Top Features:

Blood pressure (ap_hi, ap_lo).

Age (age_years).

BMI and obesity status.

🌐 Deployment The application is deployed on Streamlit Cloud: 🔗 Live Demo

App Workflow:

User inputs health data.

Real-time calculation of BMI and pulse pressure.

Model predicts risk using preprocessed features.